Extending the SMORE framework at both dataset and operator levels

[ZHANGWENTAI]

Background

SMORE has 6 algorithms for knowledge graph reasoning so far. However, it still has some limitations. For example, it limits in using type information of triple-based KG, and reasoning on Temporal KG and Hyper-relational KG.

Goal

We are going to extend SMORE to support three more algorithms in different categories:

• TeMP, which can levarage the type information of Knowledge Graph;
• TFLEX on Temporal Knowledge Graph;
• NQE on Hyper-relational Knowledge Graph

In detail, we plan to integrate different types of knowledge graph datasets, and different operators.

TODOs

• TeMP
• TFLEX
• NQE

References

[1] Hu, Zhiwei, et al. "Type-aware embeddings for multi-hop reasoning over knowledge graphs." arXiv preprint arXiv:2205.00782 (2022).

[2] Lin, Xueyuan, et al. "TFLEX: Temporal Feature-Logic Embedding Framework for Complex Reasoning over Temporal Knowledge Graph." arXiv preprint arXiv:2205.14307 (2022).

[3] Luo, Haoran, et al. "Nqe: N-ary query embedding for complex query answering over hyper-relational knowledge graphs." Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 37. No. 4. 2023.

[Hanjun-Dai]

Hi Wentai,

Thank you so much for sharing this proposal!
We look forward to your contributions, and please let us know if you encounter any issue during the development.

Best,
Hanjun